1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
|
package de.lmu.ifi.dbs.elki.utilities.scaling.outlier;
/*
This file is part of ELKI:
Environment for Developing KDD-Applications Supported by Index-Structures
Copyright (C) 2015
Ludwig-Maximilians-Universität München
Lehr- und Forschungseinheit für Datenbanksysteme
ELKI Development Team
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
import de.lmu.ifi.dbs.elki.database.ids.DBIDIter;
import de.lmu.ifi.dbs.elki.database.relation.DoubleRelation;
import de.lmu.ifi.dbs.elki.math.DoubleMinMax;
import de.lmu.ifi.dbs.elki.math.MeanVariance;
import de.lmu.ifi.dbs.elki.math.statistics.distribution.GammaDistribution;
import de.lmu.ifi.dbs.elki.result.outlier.OutlierResult;
import de.lmu.ifi.dbs.elki.utilities.documentation.Reference;
import de.lmu.ifi.dbs.elki.utilities.optionhandling.AbstractParameterizer;
/**
* Scaling that can map arbitrary values to a probability in the range of [0:1],
* by assuming a Gamma distribution on the data and evaluating the Gamma CDF.
*
* Reference:
* <p>
* H.-P. Kriegel, P. Kröger, E. Schubert, A. Zimek<br />
* Interpreting and Unifying Outlier Scores<br />
* Proc. 11th SIAM International Conference on Data Mining (SDM), Mesa, AZ, 2011
* </p>
*
* @author Erich Schubert
*/
@Reference(authors = "H.-P. Kriegel, P. Kröger, E. Schubert, A. Zimek", //
title = "Interpreting and Unifying Outlier Scores", //
booktitle = "Proc. 11th SIAM International Conference on Data Mining (SDM), Mesa, AZ, 2011", //
url = "http://dx.doi.org/10.1137/1.9781611972818.2")
public class MinusLogGammaScaling extends OutlierGammaScaling {
/**
* Maximum value seen
*/
double max = 0;
/**
* Minimum value (after log step, so maximum again)
*/
double mlogmax;
/**
* Constructor.
*/
public MinusLogGammaScaling() {
super(false);
}
@Override
protected double preScale(double score) {
assert (max > 0) : "prepare() was not run prior to using the scaling function.";
return -Math.log(score / max) / mlogmax;
}
@Override
public void prepare(OutlierResult or) {
meta = or.getOutlierMeta();
// Determine Minimum and Maximum.
DoubleMinMax mm = new DoubleMinMax();
DoubleRelation scores = or.getScores();
for(DBIDIter id = scores.iterDBIDs(); id.valid(); id.advance()) {
double score = scores.doubleValue(id);
if(!Double.isNaN(score) && !Double.isInfinite(score)) {
mm.put(score);
}
}
max = mm.getMax();
mlogmax = -Math.log(mm.getMin() / max);
// with the prescaling, do Gamma Scaling.
MeanVariance mv = new MeanVariance();
for(DBIDIter id = scores.iterDBIDs(); id.valid(); id.advance()) {
double score = scores.doubleValue(id);
score = preScale(score);
if(!Double.isNaN(score) && !Double.isInfinite(score)) {
mv.put(score);
}
}
final double mean = mv.getMean();
final double var = mv.getSampleVariance();
k = (mean * mean) / var;
theta = var / mean;
atmean = GammaDistribution.regularizedGammaP(k, mean / theta);
// logger.warning("Mean:"+mean+" Var:"+var+" Theta: "+theta+" k: "+k+" valatmean"+atmean);
}
/**
* Parameterization class.
*
* @author Erich Schubert
*
* @apiviz.exclude
*/
public static class Parameterizer extends AbstractParameterizer {
@Override
protected MinusLogGammaScaling makeInstance() {
return new MinusLogGammaScaling();
}
}
}
|
